MU: The Gold Rush Is Building More Mines
July 11, 2026 | AI demand created a memory shortage. The resulting profits are now financing the capacity that could end it.
Rating: Watch
1–3 months: The trend is cooling after an extraordinary run.
12+ months: The business has changed. The durability of its economics remains untested.
Micron’s fiscal third quarter produced $41.46 billion of revenue, up 74% sequentially, with an 84.9% non-GAAP gross margin.
The headline was AI memory. The income statement was scarcity.
Micron’s prepared remarks show why. DRAM revenue rose 67% sequentially, even though bit shipments increased only in the low-single digits. Average prices rose in the low-60% range.
NAND was even more extreme: revenue rose 99%, against mid-single-digit shipment growth, because average prices increased in the mid-80% range.
Volume moved. Price did the heavy lifting.
HBM helps explain the gap. AI accelerators require enormous amounts of high-bandwidth memory, which consumes more wafer capacity per bit than conventional DRAM. Producers direct scarce capacity toward the higher-value product, tightening ordinary memory too.
The effect reached every business unit. Gross margins ranged from 79% to 87%, including 87% in Mobile and Client. This was not just an HBM result. Scarcity repriced memory across servers, phones, PCs and cars.
Sixteen contracts meet one old cycle
Micron is trying to protect those economics before supply returns.
Its 16 strategic customer agreements cover roughly 20% of DRAM volume and one-third of NAND volume separately. Fourteen represent approximately $100 billion of minimum-price × minimum-volume contractual value. Some include fixed prices or floor-and-ceiling bands; customers commit to minimum volumes.
That improves visibility. It does not prove durability.
The agreements have never faced a memory downturn. Their real test comes when spot prices fall beneath the contractual floor: do customers honor the terms, renegotiate them or find another supplier?
An 85% gross margin earned during a shortage cannot answer that question.
The gold rush builds more mines
Micron expects DRAM and NAND conditions to remain tight beyond calendar 2027. It is also spending aggressively to increase supply.
Fiscal 2026 capital spending should reach approximately $27 billion. Fiscal Q4 alone should approach $10 billion, with quarterly spending expected to rise again in fiscal 2027. New wafer output in Idaho and Taiwan begins arriving in 2027. Samsung and SK Hynix are expanding too.
The shortage is financing the supply response that may eventually end it.
Contracts could make the next downturn shallower. They cannot stop additional capacity from testing prices. Until spot pricing weakens, both the structural bull case and the cyclical bear case remain partly untested.
Two clocks for the stock
Micron closed Friday at $979.30, roughly 22% below its June high. The long-term trend remains positive, but the stock has spent seven sessions below its 20-day average. A reclaim of roughly $991 would improve the near-term setup; losing the 50-day average near $899 would weaken it.
Fiscal Q4 is the nearer business test: $50 billion of revenue, an approximately 86% gross margin and about $31 of non-GAAP EPS. Another beat would confirm that scarcity remains powerful. It would not establish what Micron earns after supply catches up.
The decisive evidence arrives with the first meaningful decline in memory pricing. If margins hold because contractual floors work, Micron may deserve a different valuation framework. If margins follow spot prices down, the old cycle survives.
The four-line test
When a cyclical company reports explosive growth, separate:
Volume: How much more product was sold?
Price: How much growth came from scarcity?
Mix: Did better products improve the economics?
Durability: What protects those economics when supply returns?
Micron scores brilliantly on price and mix. Volume grew much less. Durability is the open exam.
The AI story is proven. The pricing story still has to survive its own success.
Subscribe for free to receive new posts.
Disclaimer: I have no crystal ball, no DeLorean, and no sports almanac from the future. This is research, not financial advice or a recommendation.